Abstract
Background:
Variations in alertness and attention are common in Lewy body diseases (LBD) and among the core features of dementia with Lewy bodies (DLB). Dopamine transporter SPECT is an accurate biomarker of nigrostriatal degeneration (NSD) in LBD.
Objective:
The present study investigated performance on a computerized alertness test as a potential measure of attention in patients with NSD compared to patients without NSD.
Methods:
Thirty-six patients with cognitive impairment plus at least one core feature of DLB referred for [123I]FP-CIT SPECT imaging were prospectively recruited. Performance in a computerized test of intrinsic alertness was compared between patients with and those without NSD as assessed by [123I]FP-CIT SPECT.
Results:
Reaction times to auditory stimuli (adjusted for age, sex, and education) were significantly longer in patients with NSD compared to those with a normal [123I]FP-CIT SPECT scan (p < 0.05). Statistical analyses revealed no significant differences comparing reaction times to visual stimuli or dispersion of reaction times between groups. Exploratory analysis in a subgroup of patients with available [18F]FDG PET revealed that longer reaction times were associated with decreased glucose metabolism in the prefrontal cortex (statistical parametric mapping, adjusted for age and sex; p < 0.005, cluster extent > 50 voxels).
Conclusion:
Computerized assessment of auditory reaction times is able to detect alertness deficits in patients with NSD and might help to measure alertness deficits in patients with LBD and NSD. Future studies in larger samples are needed to evaluate the diagnostic utility of computerized alertness assessment for the differential diagnosis of LBD.
INTRODUCTION
Early and correct diagnosis of dementia with Lewy bodies (DLB) is of prognostic and therapeutic importance, given a median survival of 5 years from symptom onset [1]. Moreover, inaccurate differential diagnosis often leads to an inappropriate prescription of antipsychotics for these patients, whereby approximately 40% of DLB patients show neuroleptic hypersensitivity [2], which implies an increased incidence of akinetic crisis as a possibly lethal event [3]. However, DLB is clinically under-diagnosed, with a sensitivity of the clinical diagnosis as low as 32% [4], and the lack of a clear definition of fluctuating alertness as one of the core criteria impairs detection of DLB.
Variations in alertness and attention are among the core features of DLB [5, 6]. Typically, these fluctuations appear ‘delirium-like’ with spontaneous alterations in cognition, attention, and arousal: including waxing and waning episodes of behavioral inconsistency, incoherent speech, and episodic staring or zoning out [5]. Yet, how these fluctuations should be assessed in clinical routine is not well defined. Previous attempts to study fluctuations comprised mainly caregiver reports and observer rating scales. Direct questioning of an informant about fluctuations may not reliably discriminate between DLB and Alzheimer’s disease (AD), though questions about daytime drowsiness, lethargy, staring into space, or episodes of disorganized speech seem to discriminate. Accordingly, these have been incorporated into scales that either score the severity and frequency of fluctuations derived from a clinical interview or use informant reports from semi-structured questionnaires [7–9]. However, assessment methods relying on expert judgement often show poor inter-rater reliability [9]. Reported agreement between raters was low to very low (58% and Cohen’s kappa = 0.25 [10]; kappa = 0.06 [11]). In this light, there is a need for standardized, objective tests for fluctuating attention in dementia patients. Given that fluctuations in DLB patients typically occur on a second-to-second basis [9], while in AD patients variations on a day-to-day basis have been described [7], a brief recording of variations in attentional performance using repeated computer-based tests may offer an independent method. Such a test typically provides not only information about variations of alertness, but also simple reaction time, which might be specifically impaired in DLB [12].
Dopamine transporter single-photon emission computed tomography (SPECT) is able to detect nigrostriatal degeneration (NSD), a typical feature of Lewy body diseases (LBD), and hence an accurate biomarker of LBD with 86% accuracy, 80% sensitivity, and 92% specificity in an study with autopsy-confirmed cases [13].
Against this background, the present study investigated whether performance in a computerized alertness test was impaired in patients with NSD. As detection of alertness deficits might be of particular value in the differential diagnosis of DLB, the present study entailed the inclusion of all patients with cognitive impairment and at least one core feature of DLB referred for SPECT imaging in clinical routine. This led to a clinical sample comprising mainly DLB patients, but also patients with Parkin-son’s disease and mild cognitive impairment (PD-MCI) or PD with dementia (PDD), which are currently considered as subtypes of the α-synuclein-associated disease spectrum of LBD [11]. To this end, we compared the performance of patients with NSD on [123I]FP-CIT SPECT, to that of patients without NSD (i.e., normal SPECT scan). We speculated to find mainly larger variations in alertness but also impaired reaction times in patients with NSD. Finally, we explored neuronal correlates of attention deficits using [18F]fluorodeoxyglucose positron emission tomography ([18F]FDG PET) as a marker of neuronal activity and neurodegeneration.
MATERIALS AND METHODS
Patients
The local ethics committee approved all procedures (proposal no. 402/17), which have therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. Thirty-six consecutive out-patients from the Memory Clinic of the Center of Geriatrics and Gerontology Freiburg (tertiary referral center) were prospectively included in the study if they 1) had cognitive impairment, 2) showed at least one clinical core feature of DLB according to consensus criteria described by McKeith et al. [6], and 3) were referred for diagnostic [123I]FP-CIT SPECT imaging. The consensus criteria of DLB from 2005 were applied as these were current at the time of the ethics committee approval. Nine of the 36 patients suffered from a tremor of the dominant hand. Twenty-seven out of 36 patients (75%) had received further diagnostic imaging using [18F]FDG PET.
Alertness tests
A neuropsychologist experienced in dementia diagnostics performed a computerized test of intrinsic alertness, which is part of a test battery to assess sensory and attention functions (WAF) [14]. This test battery contains subtests for different attentional functions, such as intrinsic alertness, and is regularly used by psychologists and neuropsychologists to assess attentional deficits in healthy participants or in neurological patients. The two subtests ‘auditory alertness’ and ‘visual alertness’ (WAFA short form) were administered each for two minutes, after instruction and a training period. Button presses of the dominant hand (right hand in all cases) in response to a sound or visually presented square on the computer display and reaction times were recorded, respectively. The fluctuation questionnaires Clinician Assessment of Fluctuation (CAF) [15] and One Day Fluctuation Assessment (ODFAS) [15] were completed with relatives as informants.
SPECT and PET imaging
[123I]FP-CIT SPECT scans were acquired on a dual-headed SPECT system (E.CAM, Siemens, Germany) equipped with a low-energy high-resolution parallel-hole collimator starting at 3 h after intravenous bolus injection of 194±7 MBq [123I]FP-CIT (GE Healthcare, Germany; thyroid uptake blocked with sodium perchlorate). The average delay between alertness testing and [123I]FP-CIT SPECT was 96 days (S.D.=158 days). Alertness testing and [123I]FP-CIT SPECT were performed within one months in 22/36 patients and within one year in 32/36 patients. For the remaining 4 patients, the delay was 372, 393, 534, and 566 days, respectively. For acquisition and processing protocols please see [16]. Visual readings of [123I]FP-CIT SPECT were performed by a nuclear medicine expert and complemented by a semi-quantitative evaluation. [123I]FP-CIT SPECT findings were dichotomized into positive or negative for nigrostriatal degeneration.
For [18F]FDG PET acquisition and processing protocols, please see [17]. All patients underwent [18F]FDG PET examinations on a Philips Gemini TrueFlight 64 integrated PET/CT system (10 min-scan, starting 50 min after bolus injection of 215±9 MBq [18F]FDG; 3D-RAMLA reconstruction). The median interval between alertness test and [18F]FDG PET scan was 27 days (more than 6 month in 8 patients).
Statistics
WAFA performance measures (reaction times and variability of reaction times from the auditory and the visual subtests) were adjusted for age, sex, and education. WAFA measures variability of reaction times in terms of the standard deviation of the log-transformed reaction times (termed ‘dispersion’). Adjusted and z-transformed WAFA performance was compared between groups using analysis of covariance (ANCOVA), controlling for hand tremor. Questionnaire scores were compared between groups using the Mann-Whitney U test. Demographic variables were compared employing either t-test or chi-squared test.
Statistical parametric mapping of [18F]FDG PET data
Individual [18F]FDG PET scans were spatially normalized to an in-house [18F]FDG template in MNI space, smoothed with a Gaussian filter of 12 mm FWHM, proportionally scaled to total brain parenchyma and subjected to a regression analysis with adjusted WAFA reaction time as predictor and co-variables age and sex (n = 27 available scans). We investigated the negative association between reaction times and [18F]FDG uptake at a predefined exploratory threshold of p < 0.005 and a minimum cluster extent of 50 voxels (1.35 mL).
RESULTS
Patient characteristics
Neuropsychiatric assessment of n = 36 patients with cognitive impairment revealed the presence of one (n = 18), two (n = 13), or three (n = 5) of the core criteria for DLB [6] (Supplementary Table 1). Eighteen patients showed NSD on [123I]FP-CIT SPECT, and 18 patients showed normal [123I]FP-CIT SPECT scans. At the time of alertness testing, there were no significant differences between the groups of patients with and without NSD in terms of sex, age, educational years, Mini-Mental State Examination (MMSE), symptom duration or medication history concerning dopaminergics and anti-dementia drugs. The average Unified Parkinson’s Disease Rating Scale motor score was significantly higher in the NSD group compared to the group of patients without NSD (p = 0.005). The former group also showed significantly higher Hoehn and Yahr stages than the latter group (p = 0.02) (Table 1).
Demographic characteristics of patients
NSD, patients with nigrostriatal degeneration on FP-CIT SPECT; TMT-A, trail-making test, part A. #UPDRS III was available in 35 out of 36 patients. Differences at group level (t-test): *p = 0.095; **p = 0.02; §p = 0.005.
Performance measures of alertness tests
ANCOVA revealed significantly prolonged reaction times to auditory stimuli in the NSD group compared to the group of patients without NSD (F(1, 33)= 5.1, p = 0.030, Cohen’s d = 0.78; Fig. 1). This group difference was essentially replicated when controlling for medication with dopaminergics or anti-dementia drugs. No other performance measure of auditory or visual alertness showed a significant difference between groups (Table 2). Exploratory ANCOVA using the Unified Parkinson’s Disease Rating Scale motor score instead of hand tremor as a covariate failed to show a significant difference between groups in any of the alertness test performance measures (p > 0.1). As a further exploratory analysis, we compared groups of patients with only one versus those with at least two of the DLB core criteria [5] and observed no significant group difference in any of the WAFA performance measures (all p > 0.1, ANCOVA controlling for hand tremor).
Intrinsic alertness to auditory stimuli (left) was prolonged in patients with nigrostriatal degeneration on FP-CIT-SPECT. Intrinsic alertness to visual stimuli (right) showed no group difference.
Alertness test performance raw data and questionnaire scores
For abbreviations, see Table 1.
Alertness questionnaires
No significant difference was found between the two groups (NSD versus no NSD) regarding the scores of the questionnaires ODFAS and CAF (p > 0.1, respectively; Table 2).
[18F]FDG PET imaging correlates
Longer reaction times were associated with de-creased glucose metabolism in the lateral and medial prefrontal cortex, predominantly within the right hemisphere (Fig. 2, Table 3). Longer reaction times were associated with increased metabolism mainly in the left fusiform gyrus (peak voxel coordinates -40/-32/-20, t = 5.17, p < 0.001). When adjusting for MMSE as a proxy for disease severity, the association between longer reaction times and decreased glucose metabolism in the frontal lobe was widely retained. By contrast, when adjusting for [123I]FP-CIT SPECT positivity, no association between reaction times and glucose metabolism in the frontal lobe remained.
Regions with negative association between regional glucose metabolism and reaction times to auditory stimuli (p < 0.005, cluster extent > 50 voxels): glass brain views from top, right, and front (from left to right).
Clusters of voxels showing a significant negative association between glucose metabolism and reaction times.
n = 27; p < 0.005, cluster extent > 50 voxels (1.35 mL).
DISCUSSION
The present study was undertaken to investigate performance in a computerized alertness test as a potential measure of attention deficits in patients with LBD. As the clinical diagnosis of LBD is often inaccurate, particularly for DLB [18], we tested against a biomarker with excellent diagnostic accuracy for LBD: dopamine transporter SPECT [13, 19]. Alertness to auditory stimuli was impaired in patients with NSD, compared to SPECT-negative patients, who served as a control group in the present study. It was also impaired compared to reaction times of healthy individuals as reported in the literature (259±31 ms [20]). By contrast, other performance measures of computerized alertness tests did not differ between the two patient groups, nor did scores of clinical alertness questionnaires.
Patient groups
A large clinico-pathological study revealed that the sole clinical diagnosis of PDD/DLB is notoriously inaccurate (erroneous in 65% of cases), this pertains in particular to the differentiation between PDD/DLB and AD [13]. The present study circumvents this clinical pitfall by focusing on the assessment of nigrostriatal integrity using [123I]FP-CIT-SPECT. Several studies including postmortem verified cases provide support to the notion that striatal dopamine transporter availability is a valuable measure for an early differential diagnosis of DLB in vivo with an estimated pooled sensitivity and specificity of 87% and 94%, respectively [13, 19]. In this light, reduced dopamine transporter uptake in the basal ganglia demonstrated by SPECT or PET is regarded as an indicative biomarker of DLB in current consensus criteria [5]. Thus, those patients of the present study cohort exhibiting a pathological [123I]FP-CIT SPECT scan can be considered to likely have LBD, while the remaining ones most likely suffer from another disease (n = 7 AD, n = 6 psychiatric disorder, n = 4 frontotemporal dementia, n = 1 limbic encephalitis).
Fluctuating cognition questionnaires
An early study on the development and validation of CAF (81% /92%) and ODFAS (93% /87%) yielded high sensitivity and specificity for distinguishing DLB from AD [15]. In good agreement with our findings, Bradshaw et al. [7] noted that the ODFAS only detected fluctuations in a minority of patients with DLB (46%), since the quantitative domains of ODFAS and CAF tend to occlude important discriminative qualitative differences of cognitive fluctuations between patients with AD and DLB. A study applying a cluster analysis using CAF combined with the Mayo Composite Fluctuations Scale reported no significant result for fluctuations, while visual hallucinations and motor symptoms seemed useful to distinguish DLB from other dementia [21]. Thus, a meta-analysis on the diagnostic validity of clinical alertness questionnaires concluded that psychometric measures developed for the identification and assessment of fluctuations have not been adequately tested as yet for reliability and validity [22].
Computerized alertness tests
In our view, the choice of testing particularly intrinsic alertness in this study is justified in two ways: 1) intrinsic (or endogenous) alertness, typically measured by simple reaction times without a preceding warning stimulus, represents the cognitive control of wakefulness and arousal (as opposed to phasic alertness, which is typically assessed by reaction time tests with preceding warning stimuli) [23] and is as such potentially relevant to DLB. 2) Delays in simple reaction times in DLB have been reported previously: In line with our findings, a large prospective study (n = 85 DLB, n = 80 AD, n = 35 controls) employing the Cognitive Drug Research Computerized Assessment System for Dementia Patients found that patients with DLB were significantly more impaired than patients with AD on all measures of attention and fluctuating attention [12]. For diagnostic verification the first 50 cases underwent postmortem examination. Importantly, in this study [12] impairment of cognitive reaction times was specific to DLB patients, whereas other measures were abnormal in DLB as well as in AD patients. In line with our data, this suggests that cognitive reaction times might be used for identification of DLB in the diagnostic process. Further studies [9, 24] confirmed that patients with DLB had a greater prevalence and severity of cognitive fluctuations than patients with AD or vascular dementia, when applying 90-second cognitive choice reaction time and vigilance reaction time trials.
Imaging data
To our knowledge, this study is the first to investigate neuronal correlates of alertness in patients with clinical features of DLB using [18F]FDG PET, an established method for assessing cerebral glucose metabolism and neuronal activity. The observed brain regions associated with prolonged reaction times overlap with those reported in previous task-related perfusion PET studies on intrinsic alertness: Sturm and colleagues described a predominantly right-sided network of brain regions comprising parts of the frontal and cingulate cortex [23, 25]. They further overlap with the activation reported in an fMRI study (especially regarding anterior cingulate cortex involvement) which employed a modified version of the same intrinsic alertness task used in our study [20]. The marked right-ward asymmetry of the regions we identified is in line with the aforementioned study and early reports on intrinsic alertness deficits in right hemispheric stroke patients [26].
Available imaging data in patients with fluctuations of alertness is sparse and somewhat conflicting: One previous study using fMRI suggested thalamic damage and cholinergic imbalance as potential correlates of fluctuations in DLB [27], while another group reported a correlation between reduced functional connectivity in the right hemisphere and severity of fluctuations [28]. A recent study described an association between a desynchronization of a number of cortical and subcortical areas related to the left fronto-parietal network and the severity and frequency of cognitive fluctuations [29]. The fact that our study revealed decreased prefrontal neuronal activity (right more than left) associated with longer reaction times might indicate a deficit of the fronto-parietal alerting network [30]. Moreover, glucose metabolism in the identified prefrontal region was markedly lower in those patients with a dopaminergic deficit, and—consequently—no association between reaction times and glucose metabolism in the frontal lobe remained when adjusting for effects of [123I]FP-CIT SPECT positivity. These observations might point to an underlying malfunction of the dopaminergic system.
Limitations
The number of patients in the present study is relatively small, possibly precluding the detection of statistically significant performance differences beyond those observed here (e.g., regarding reaction times to visual stimuli, which were also impaired in patients with NSD, though not statistically significantly). Performance of NSD patients was not compared to a healthy control group, but to a heterogeneous group of SPECT-negative patients. While this complicates the interpretation of group differences, it reflects the typical clinical scenario. The time gap between imaging and alertness testing was more than one year in 4/36 patients, two with NSD, two without NSD. Exclusion of these four patients did not substantially change the results (data not shown in detail). The short test duration in the current study might explain why no significant effects of reaction time variability (within patients) was observed, despite the prominent role of fluctuating alertness in current consensus criteria [5]. Although a previous study reported fluctuations of cognitive performance in DLB over the course of the day [31], our aim was to assess the utility of a neuropsychological test that could easily be added to an existing diagnostic test battery. The choice of this brief test duration was also encouraged by previous observations of alertness deficits in DLB within even shorter time spans [9]. Given that the clinical distinction between PDD and DLB is controversial, relying on the somewhat arbitrary “1-year rule” [32] it has to be noted that our sample comprised both, patients with PDD as well as DLB. However, previous studies revealed no significant differences comparing cognitive fluctuations between PDD and DLB patients [33, 34]. While striatal dopamine transporter availability is a valuable measure for an early differential diagnosis of DLB in vivo, the limited sensitivity of [123I]FP-CIT SPECT of around 80% for identification of LBD [13] points to the fact that ‘NSD’ and ‘no NSD’ in this work are not necessarily synonymous to ‘LBD’ and ‘no LBD’, respectively.
Conclusion
Computerized assessment of auditory reaction times is able to detect alertness deficits in patients with NSD and might help to measure alertness deficits in patients with LBD and NSD. Future studies in larger samples are needed to evaluate the diagnostic utility of computerized alertness assessment for the differential diagnosis of LBD.
